Data inspection, editing & Radio-Frequency Interference flagging

Similar documents
Charting the Transient Radio Sky on Sub-Second Time-Scales with LOFAR

Living with radio interference (rfi) Define Interference Please

From Visibilities to Images Tom Muxlow, JBCA

Ionospheric Research with the LOFAR Telescope

The Solar Science Data Center and LOFAR

Self-Calibration and Hybrid Mapping

The Phase Modulator In NBFM Voice Communication Systems

INTRODUCTION TO COMMUNICATION SYSTEMS AND TRANSMISSION MEDIA

A comparison of radio direction-finding technologies. Paul Denisowski, Applications Engineer Rohde & Schwarz

The human pipeline: distributed data reduction for ALMA

CASA the science data analysis package for ALMA

Spectrum and Power Measurements Using the E6474A Wireless Network Optimization Platform

ASKAP Science Data Archive: Users and Requirements CSIRO ASTRONOMY AND SPACE SCIENCE (CASS)

Interferometría... Magia o Ciencia? Rubén Herrero-Illana

Tektronix RSA306 USB Spectrum Analyzer

Analysis of ATCA Radio Continuum Data towards Imaging a Radio Galaxy

EA-ARC ALMA ARCHIVE DATA USER GUIDEBOOK

UV Continuum Subtraction Tasks in CASA

A User s Guide to ALMA Scheduling Blocks (Cycle 3)

SURVEYING WITH GPS. GPS has become a standard surveying technique in most surveying practices

P25 / TETRA Network Planning Using EDX SignalPro

Amy Mioduszewski NRAO. NRAO s 2008 Synthesis Imaging Workshop) Synthesis Imaging in Radio Astronomy

REQUIREMENTS TRACEABILITY

CASA the science data analysis package for ALMA

AN Application Note: FCC Regulations for ISM Band Devices: MHz. FCC Regulations for ISM Band Devices: MHz

The CASA software package

CASA. Nemesio Rodriguez-Fernandez (IRAM) Outline. - Status and Measurement Set. - Structure. - Tools. - Tasks. - Scripts.

RF Measurements Using a Modular Digitizer

Planning Terrestrial Radio Networks

K2 CW Filter Alignment Procedures Using Spectrogram 1 ver. 5 01/17/2002

DAB1001. Wireless Digital Radio Interface. Installation & User Guide

SATELLITE COMMUNICATION STATION LICENCE APPLICATION GUIDELINES

ALMA Technical Support. George J. Bendo UK ALMA Regional Centre Node University of Manchester

Astronomical applications of the over-the-horizon radar NOSTRADAMUS

The front end of the receiver performs the frequency translation, channel selection and amplification of the signal.

Synthetic Sensing: Proximity / Distance Sensors

GSM frequency planning

Application Note. EtherCAT Master Architecture. Applicable Products: Yaskawa Servodrives with CANopen over EtherCAT

Dynamic Frequency Selection (DFS) and the 5GHz Unlicensed Band

Data Analysis Methods: Net Station 4.1 By Peter Molfese

AM TRANSMITTERS & RECEIVERS

Information Paper. FDMA and TDMA Narrowband Digital Systems

Spectrum Analyzers vs. Monitoring Receivers. Paul Denisowski, Application Engineer Rohde & Schwarz

VANGUARD ONLINE BACKUP

Radio-frequency channel arrangements for high-capacity fixed wireless systems operating in the lower 6 GHz (5 925 to MHz) band

Agilent Creating Multi-tone Signals With the N7509A Waveform Generation Toolbox. Application Note

Doppler. Doppler. Doppler shift. Doppler Frequency. Doppler shift. Doppler shift. Chapter 19

Module 13 : Measurements on Fiber Optic Systems

Command-induced Tracking Jitter Study I D. Clark November 24, 2009

AMS DAQ, TRD DSP Software

FEATURE GUIDE AeroSphère / Small / Large / Base

Big bang, red shift and doppler effect

Revision of Lecture Eighteen

The Murchison Widefield Array Data Archive System. Chen Wu Int l Centre for Radio Astronomy Research The University of Western Australia

OSKAR Example Revision: 8

RECOMMENDATION ITU-R SM Measuring sideband emissions of T-DAB and DVB-T transmitters for monitoring purposes

Preview of Period 3: Electromagnetic Waves Radiant Energy II

Impedance 50 (75 connectors via adapters)

Attenuation (amplitude of the wave loses strength thereby the signal power) Refraction Reflection Shadowing Scattering Diffraction

Network analyzer and spectrum analyzer two in one

Evolution of Satellite Communication Systems

Introduction to RF Engineering. Andrew CLEGG

The PACS Software System. (A high level overview) Prepared by : E. Wieprecht, J.Schreiber, U.Klaas November, Issue 1.

SR2000 FREQUENCY MONITOR

A brief introduction to the Measurement Set and the ALMA Science Data Model

Spectrum occupancy measurement and evaluation

5. Tutorial. Starting FlashCut CNC

Department of Electrical and Computer Engineering Ben-Gurion University of the Negev. LAB 1 - Introduction to USRP

Spectrum Analyzers And Network Analyzers. The Whats, Whys and Hows...

Electromagnetic radiation exposure: assessment against ACA mandated limits

Broadband over Power Line (BPL) Test Procedures and Equipment Authorization

DT3: RF On/Off Remote Control Technology. Rodney Singleton Joe Larsen Luis Garcia Rafael Ocampo Mike Moulton Eric Hatch

Online Filtering for Radar Detection of Meteors

Frequency Response of Filters

APSYN420A/B Specification GHz Low Phase Noise Synthesizer

Tx/Rx A high-performance FM receiver for audio and digital applicatons

DDX 7000 & Digital Partial Discharge Detectors FEATURES APPLICATIONS

EE4367 Telecom. Switching & Transmission. Prof. Murat Torlak

NETWORK LICENCE (Localised Private Network Multi-Channel Radiocommunication)

The role of spectrum monitoring in support of inspections

Guide for Performing a Wireless Site Survey. 2.4 GHz IEEE g/802.11b/

Network of European Research Infrastructures for Earthquake Risk Assessment and Mitigation. Report

Universal Form-factor. Wi Fi Troubleshooting Made Easy

An overview of the IEEE Standard

Distributed Synchronization

Maximizing Receiver Dynamic Range for Spectrum Monitoring

Spectral Line II. G ij (t) are calibrated as in chapter 5. To calibrated B ij (ν), observe a bright source that is known to be spectrally flat

Cisco Cable Radio Frequency (RF) FAQs

Jabra CRUISER2. User manual. MUTE VOL - VOL + jabra

Simultaneous Heliospheric Imager and Interplanetary Scintillation observations of CMEs and CIRs

MANUAL FOR RX700 LR and NR

Transcription:

Data inspection, editing & Radio-Frequency Interference flagging ERIS 2015, Garching / München André Offringa (ASTRON Netherlands Institute for Radio Astronomy) 2015-09-06

Outline Introduction: Why data editing Radio Frequency Interference (RFI) Plotting data (CASA, aoqplot) Manual data flagging Automatic RFI flagging algorithms Data averaging Sources of imaging errors

Introduction Why data editing?

Introduction Why data editing? Some antennas might not have been functioning properly...

Introduction Why data editing? Broken elements (antennas/stations) Correlator malfunctions Shadowing Initial pointing delay Bandpass issues Low elevation Correlated noise on some baselines (e.g. LOFAR split stations) Interference

Introduction Why data editing? Broken elements remove antennas Correlator malfunctions remove timesteps Shadowing remove antennas in time range Initial pointing delay remove first timesteps Bandpass issues remove channels Low elevation remove antennas with low elevation Correlated noise on some baselines (e.g. LOFAR split stations) Flag baselines Interference remove antennas, timestep, frequencies or baselines...

Data can't be (self-)calibrated when any of these issues are still in the data. Therefore, data inspection & editing is the first step : INSPECTION + EDITING (DATA AVERAGING) CALIBRATION IMAGING

First step in data reduction: Data inspection (example of casaplotms on other screen) Start casaplotms Open MS ('3c196_spw5_sub1.ms') Press 'plot' (plots amplitude vs time) Goto 'axes', select frequency as x-axis RFI is visible Select 'antenna1' antenna5 has no data (Enter: antenna: 0;13 msselect: ANTENNA1!=ANTENNA2 )

casaplotms

casaplotms What should we see in casaplotms (time vs amplitude) if we observe a single unresolved (=delta function) source with a certain flux? (That's what we want calibrators to be strong (i.e. dominating / 'single') and unresolved)

casaplotms casaplotms is useful for many things: Browsing for bad antennas, frequencies, etc. Also useful for inspecting calibration results Or getting an idea of model data Further discussed in Andy Biggs' tutorial Many observatories have specialized plotting tools

Removing data If an issue is found (bad antenna, baseline, channels,...) how do we remove it from our dataset? We don't actually remove data, we 'flag' data and ignore these in further processing. Flagging is not the same as setting to zero(!) 'taql' is a useful tool for data editing.

TaQL (Table Query Language) TaQL is an 'SQL'-like language for quick data editing of CASA data. Command line tool 'taql' available, easy for scripting Be careful when editing! Always keep backups. Some examples: (from the cmdline) taql select unique TIME from obs.ms taql update obs.ms set FLAG=true where ANTENNA1==ANTENNA2 See taql doc ( casacore note 199 ) http://www.astron.nl/casacore/trunk/casacore/doc/notes/199.html

Other ways to edit data... CASA task 'flagdata' Andy Biggs' tutorial Or, for CASA data: Write Python scripts Other packages have their own scripting languages / tasks

Radio-Frequency Interference Our radio spectrum is almost entirely allocated to services other than radio astronomy FM, airplane communication, satellite downlink, remote controls, digital broadcasts, Also accidental and natural occurrence of RFI: Cars, electrical fences, high-voltage lines (anything that sparks), lightning, the sun, etc. RFI can cause (self-)calibration to fail and/or reduce imaging sensitivity

Example of LOFAR data with RFI Cross-correlations of two stations showing strong RFI

RFI Lots of interference at low frequencies (<1.5 GHz, e.g. LOFAR, GMRT, WSRT, EVLA, MWA, ) Less of an issue for higher frequencies (ALMA); or VLBI but mitigation still required in most cases.

Excising RFI Detection methods are common in radio astronomy Common methods: Manual selection by data reducing astronomer Thresholding / specialized project pipelines (e.g. Baan et al. 2004, Winkel et al. 2007) Manual selection is not practical for modern observatories: Enormous data volumes, computationally fast algorithms required. Needs to be more accurate than thresholding

RFI stages / strategies Many RFI excision options: Online pre-/post-correlation mitigation Memory/computational constraints Required for coherent (high time res) modes Offline mitigation Post-optimizable, not real-time, data can be reordered LOFAR: Station level spatial filtering Expensive, low SNR, only one chance Allows data recovery

Automated excision of RFI Two classes of RFI excision methods: Detection: find & throw away affected data Filtering or subtracting: estimate RFI contribution and restore affected data Detection methods ( flagging ) commonly used Some specialized pipelines for surveys or instruments Filtering RFI is harder Resulting data quality is not well understood Requires more resources Lack of full (automated) filtering pipelines

The AOFlagger (example of automated RFI detection) External package1, works with CASA sets - Offringa et al., MNRAS (2010), Offringa et al., A&A (2012) 1 AOFlagger can be downloaded from http://aoflagger.sourceforge.net/

Subtracted background High-frequency components

XX cross-correlation XY cross-correlation YX cross-correlation YY cross-correlation

What could go wrong??

What could go wrong? - Some astronomical sources vary quickly in time (sun, pulsars, ) - Quick fringes are line-like patterns - Spectral line observations Mostly not an issue sources are mostly much weaker than RFI, and invisible in single correlations.

Accuracy & speed Flaggers have to be accurate, but also fast MAD flagger (comparable to Pyflag, Miriad s flagger, AIPS flagger) AOFlagger

WSRT data example

WSRT data example

Thresholding vs. AOFlagger MWA 3 min observation with 32 tiles Thresholding AOFlagger Image credit: Natasha Hurley-Walker (MWA data)

More about AOFlagger (Almost) same algorithm can be used for many telescopes Software has been successfully used for: LOFAR (Offringa et al 2012), MWA (Offringa et al. 2015), WSRT, JVLA, GMRT, ATCA, Parkes, Arecibo, and BIGHORNS I don't know if it is used for ALMA... For Miriad users: Miriad has an implementation of AOFlagger SumThreshold algorithm available in E-Merlin SERPent pipeline (Peck & Fenech, 2013)

RFI excision for LOFAR LOFAR s case: Fully automated detection, only a few % lost data Only small residuals, do not affect image quality Why such good results? LOFAR has very high time/freq resolutions Design has accounted for interference High accuracy of algorithms Some transmitters do remain problematic (e.g., DAB, FM, wind turbines) Tweaking still required for special cases

Analysing RFI (Demo: open rfigui in other window) Open set, goto RT1 x RT2. Execute strategy Edit strategy: change flagged polarizations, change sumthreshold sensitivity Save strategy Execute 'aoflagger' on cmdline.

Further analyses (Demo: open aoqplot in other window) casaplotms is slow for very big files aoqplot can give a quick overview

Always flag (first) at highest possible resolution: Highest resolution: Averaged without RFI detection: Flagging is incremental: don't reset flags! (e.g. taql update obs.ms set FLAG=false) Correlator might have set flags. These will be lost. To undo flagging, use backup (column).

Averaging & smearing Data size can be reduced by averaging data in time and/or frequency direction Only average after RFI detection Over-averaging causes smearing Time-smearing: in tangential direction Frequency-smearing: in radial direction Calibration might also constrain averaging factor Next talk by George Heald

Bandwidth smearing Off-axis sources fringe faster ( Neal Jackson's lecture) Smearing is proportional to distance from phase centre (slide by Tom Muxlow)

Averaging in uv-space --> v bandwidth time --> u

Smearing General rule: phase turn along time / frequency should be sampled << 1/4th of a turn. Example with 1'' resolution (e.g. LOFAR international baseline) and 1 deg off-axis source: Source is 3600 resolution elements away Phase turns ~3600 times in 6 hours (or over observing frequency) Need ~14000 samples in 6 hours Time res Δt < ~2 s (Δν < ~10 khz @ 150 MHz).

Data averaging with CASA (Demo: casa split) Example: (from casapy shell) inp split vis='3c196_spw5_sub1.ms' (input) outputvis='averaged.ms' width=8 (Average over 8 channels) timebin='60s' (Average over 60 s) go

Original resolution: After averaging:

Averaging DATA Processing data can be very time expensive, but almost all steps scale linear with nr. of visibilities. Work on averaged data (and/or subset) while experimenting with settings

NDPPP: Averaging LOFAR data Almost all telescopes have existing sets of scripts to do preprocessing Use them! 'split' task does not work well on LOFAR data (see LOFAR cookbook for details) Instead, a specialized LOFAR pipeline was made to perform several steps at once NDPPP: New Default Pre-processing Pipeline Can run aoflagger and perform averaging at once (as well as several other things) See LOFAR Cookbook for detailed info (MWA has a similar pipeline called 'cotter').

Averaging DATA Processing data can be very expensive, but almost all steps scale linear with nr. of visibilities. While experimenting with settings, work on averaged data

Summary First step in data processing is data inspection Second step is data flagging...or isn't it?

Summary First step in data processing is data inspection Second step is data flagging Second step is BACKUP YOUR DATA Third step is data flagging and RFI detection Calibration, imaging, to be discussed!

Summary I've shown: Data inspection (with e.g. CASA casaplotms, rfigui and aoqplot) Flagging data manually (with taql) Automated RFI detection (with the AOFlagger) Data averaging (with CASA split or NDPPP) Issues with insufficient resolution (smearing, bad RFI detection) Good luck!